Q&A anticipation

Predict the hardest questions on your work and prepare answers, so viva, defence, and audience Q&A stop being pure ambush.

When to use this: When checking yourself
Learning impact: Reliable gain
Learning evidence: Credible with caveats
Where you are with the topic: Checking readiness
How often to use it: Worth repeating often

When to use it

A viva, defence, or audience Q&A is coming and you have only polished the talk.
You can present smoothly and freeze on an obvious methods probe.
You want to practise answers that are brief, honest about limits, and on-question.

What to ask AI

Type something like: “My talk/paper claims: [ ]. Generate 12 hard questions (label types). Do NOT answer them. I will answer one by one; then critique brevity, honesty about limits, and whether I addressed the question.”

Ask for questions only first, answers must be yours.

Keep the hard part on your side: you answer under pressure.

How it helps

Q&A anticipation means predicting the hardest questions and preparing answers so viva and audience Q&A stop being ambushes.

A polished talk with deer-in-headlights on an obvious methods probe is the classic miss.

Retrieval under expected interrogation is the training; reading AI-written answers is not.

How to run it

State your core claims and known weak points.
Ask for a bank of hard questions labelled by type (methods, limits, alternatives, implications).
Answer one by one aloud or in writing without looking up crib notes.
Ask for critique on brevity, honesty about limits, and whether you answered the question asked.
Build a short sheet of your best answers in your own words.

Example exchange

You

My claims are that tutoring raised mid-year scores, and that we do not claim causation. Generate hard audience questions by type. Do not answer them for me.

Start with methods: how did you handle selection into tutoring? Answer that one first. We will add limit and alternative questions after.

You

Schools opted in. We matched on prior scores, but we cannot rule out motivation differences.

Honest answer. Tighten it to one sentence on the match and one on residual bias. Next question: what else could explain the bump besides tutoring?

Copyable prompt

My talk/paper claims: [ ].
Generate 12 hard questions (label types). Do NOT answer them. I will answer one by one;
then critique brevity, honesty about limits, and whether I actually addressed the question.

The Tell

A polished talk with deer-in-headlights on an obvious methods probe means you rehearsed presentation, not defence.

If you only memorised AI-written answers, the first unexpected follow-up will still break you.

Principle evidence

Strength of the underlying learning idea, not a claim about AI products.

The underlying learning idea is rated moderate. Anticipating hard questions is retrieval practice under expected interrogation, kin to viva preparation and formative oral assessment.

AI delivery evidence

Whether an AI tutor delivers this method well is a separate question.

AI question generation is useful and uneven; field-specific killers may be missing. Your spoken answers are the learning event. Delivery evidence for AI Q&A coaches is speculative.